Heart

Menu

Stay in the loop

Never miss a beat with the latest in VET news delivered straight to your inbox.

Stay in the loop

Never miss a beat with the latest in VET news delivered straight to your inbox.

Enhanced Student Success Through Chatbots

November 13, 2017

After-hours tutoring is a great way to consolidate a student’s understanding of their learning material, and ensure your participants are getting the most out of their course. If your students are turning to external tutoring sources, consider providing your own services to help them with their studies.

The most cost-effective method of providing extra study help is to utilise a chatbot — and while this may sound daunting, AI integration is becoming more commonplace and more accessible. Here are 3 tips to help you get started:

1. Text data collection

This process can be really easy if any conversational data set has already been made available to you, otherwise, you may have to make that happen by yourself. One option is to create a teacher-student online communication channel, and hire staff to provide textual responses to the students. Theoretically, the conversation can also take place in voice or video chats, or through voice recognition techniques. The advantage of collecting data on your own is that the training data will be more relevant, resulting in high quality bot conversations.

2. Text pre-processing

When the conversation data has accumulated to a satisfactory amount, the data can be applied to build the chatbot for tutoring purposes. Before it’s fed as input into a model, the text needs to go through pre-processing: stop word removal, punctuation removal, lemmatisation, and tokenisation. This is essential to ensure that word variations, regardless of tense, singular or plural forms, will always be recognised as the same word.

3. Language model

The term chatbot doesn’t actually refer to a robot that sits on the other end of the connection and type out answers on a keyboard, although that’s how most people picture it. A chatbot is really just a model that predicts the possibility of words, aka a language model. By applying recurrent neural networks, the model takes a sequence of words in time order, trains its parameters to memorise the training text data, and then predicts the words for response when given the question.

By utilising this technology, VET organisations may customise their very own tutoring chatbots to ensure their learners are getting the most out of their training.